Monitoring the performance of hydraulic pumping equipment
Mixing equipment for hydraulic fracturing operations undergoes wear with continued use. Monitoring the performance of the mixing equipment and predicting failure may be performed by measuring discharge pressures or hydrodynamic acoustic noise. Computer modeling and machine learning may be employed to help operators decide whether to take equipment out of service for maintenance or to allow the equipment to continue operating.
1. A method, comprising:
(a) measuring a first discharge pressure versus revolutions per minute (RPM) in a first vortex mixer;
(b) determining a condition of a second vortex mixer based on, at least in part, a second discharge pressure versus RPM in the second vortex mixer;
(c) estimating a discharge pressure drift rate versus RPM;
(d) predicting an expected time when the first vortex mixer will reach the condition of the second vortex mixer, wherein the expected time is predicted based on the first discharge pressure versus RPM, the second discharge pressure versus RPM, and the discharge pressure drift rate versus RPM;
(e) monitoring the first vortex mixer over a period of time prior to the expected time to determine a current condition of the first vortex mixer, wherein the current condition of the first vortex mixer is based on, at least in part, the first discharge pressure versus RPM; and
(f) removing the first vortex mixer for maintenance based on a correspondence between the current condition of the first vortex mixer and the condition of the second vortex mixer.
2. The method of claim 1 , wherein stage (c) is performed with the first vortex mixer and the second vortex mixer.
3. The method of claim 1 , wherein stage (c) is performed using laboratory measurements.
4. The method of claim 1 , wherein stage (c) is performed using a modeling approach.
5. The method of claim 4 , wherein the modeling approach comprises using software comprising ANSYS or STAR-CCM+.
6. The method of claim 1 , wherein machine learning methods are used to perform stages (a)-(d), the machine learning methods comprising linear regression models, ensemble models or neural networks or combinations thereof.
7. The method of claim 1 , wherein the first vortex mixer and the second vortex mixer are operated between 500 and 5000 RPM.
8. The method of claim 1 , wherein a hydrophone or a high-frequency pressure gauge is employed to detect hydrodynamic acoustic noise.
9. A method, comprising:
(a) measuring a first hydrodynamic acoustic noise spectra from first pumping equipment;
(b) determining a condition of second pumping equipment based on, at least in part, a second hydrodynamic acoustic noise spectra from the second pumping equipment;
(c) estimating a drift rate of the hydrodynamic acoustic noise spectra;
(d) predicting an expected time when the first hydrodynamic noise spectra will reach that of the second pumping equipment, wherein the expected time is predicted based on the first hydrodynamic noise spectra, the second hydrodynamic acoustic noise spectra, and the determined hydrodynamic noise spectra drift rate;
(e) monitoring the first pumping equipment over a period of time prior to the expected time to determine a current condition of the first pumping equipment based on, at least in part, the drift rate of the hydrodynamic acoustic noise spectra; and
(f) removing the first pumping equipment for maintenance based on a correspondence between the current condition of the first pumping equipment and the condition of the second pumping equipment.
10. The method of claim 9 , wherein stages (a)-(d) comprise monitoring the hydrodynamic acoustic noise from internal equipment surfaces.
11. The method of claim 9 , wherein stages (a)-(d) comprise monitoring the hydrodynamic acoustic noise produced by flowing fluids.
12. The method of claim 9 , wherein, during stages (a) (d), the hydrodynamic acoustic noise spectra cover a frequency range between 0 and 100 kHz.
13. The method of claim 9 , wherein stage (c) is performed with mixing or pumping equipment.
14. The method of claim 9 , wherein stage (c) is performed during laboratory experiments.
15. The method of claim 9 , wherein a modeling approach is employed while monitoring the hydrodynamic acoustic noise spectra.
16. The method of claim 15 , wherein the modeling approach comprises using software comprising ANSYS or STAR-CCM+.
17. The method of claim 9 , wherein machine learning methods are used to perform stages (a)-(d), the machine learning methods comprising linear regression models, ensemble models or neural networks or combinations thereof.
18. The method of claim 9 , wherein machine learning methods are used to perform stages (a) (d), the machine learning methods comprising linear regression models, ensemble models or neural networks or combinations thereof.
19. The method of claim 9 , wherein measuring the first hydrodynamic acoustic noise spectra from the first pumping equipment is based on, at least in part, one or more geometric parameters of the first pumping equipment.
20. The method of claim 19 , wherein the one or more geometric parameters of the first pumping equipment is determined based on sensor data from one or more sensors configured to acquire hydrodynamic noise.